Design methodology for MuGFET ESD protection devices

Abstract

A method for manufacturing a MuGFET ESD protection device having a given layout by means of a given manufacturing process, the method comprising selecting multiple interdependent layout and process parameters of which a first set are fixed by said manufacturing process and a second set are variable, selecting multiple combinations of possible layout and process parameter values which meet predetermined ESD constraints; determining an optimum value for at least one other parameter in view of a predetermined design target apart from the predetermined ESD constraints; determining values for fin width (Wfin), gate length (LG) and number of fins (N) on the basis of the optimum value; and manufacturing said MuGFET ESD protection device using the given manufacturing and process values.

Description

CROSS-REFERENCE TO RELATED APPLICATIONS[0001]The present application claims priority to European Patent Application EP 08156029.4 filed in the EPO Patent Office on May 9, 2008, the entire contents of which is incorporated herein by reference.BACKGROUND[0002]1. Field of the Invention[0003]The present invention relates to designing and manufacturing MuGFET ESD Protection Devices.[0004]2. Description of the Related Art[0005]The FinFET transistor is the most widely studied multi-gate architecture for technology-scaling below 45 nm due to its excellent control of Short Channel Effects (SCE) and its compatibility with standard CMOS processing. In Trémouilles et al, “Understanding the Optimization of Sub-45 nm MuGFET Devices for ESD Applications”, ESD 2007, pp 408-415, a variety of ESD device parameters (failure current It2, holding voltage Vh, on-resistance Ron, leakage current and area consumption) of a Grounded-Gate-NMOS FinFET device were investigated as a function of different layout ...

AI Technical Summary

Benefits of technology

This approach enables better optimization results, such as a 20% reduction in area compared to intuitive solutions, while meeting design targets and constraints, and can be easily adapted to changing design needs.

Problems solved by technology

Major challenges lie ahead for the ESD design engineer.

A design methodology is needed because the more complex geometry of FinFET devices compared to planar introduces a higher degree of complexity that make optimization absolutely not obvious as compared to planar devices.

For wide fin (planar) devices (FIG. 2 dotted lines) for small gate lengths, a VT roll-off is seen because of loss of SCE control.

Devices with too high initial leakage would consequently have a low ESD robustness, and hence are not considered.

It2 and Ron may be impacted by constraints in the metal wiring and / or accumulative heating of folded structures.

It is clear that any LG or Wfin deviation from the optimum leads to a drastic increase in the required area.

No solutions exist for small LG because of the leakage current limitation and for large LG because of too high voltage drop.

This solution is limited by the leakage current limit and results in an increased area consumption of 40%, namely 417 μm2.

On the contrary, for wide fin devices VT decreases with short gate lengths, thus these devices have poor control of SCE and hence cannot be used.

Ron-extraction in diode mode is more difficult than in bipolar mode due to the non-linear current behavior (FIG. 2).

However, the overall performance of the MOS diode is not superior to the bipolar mode, which is generally the case.

This has an important consequence, namely that in some cases, the MOS diode mode would become the limiting factor when compared to the bipolar mode.

However, when technology keeps evolving, the different dependencies might change, making the design methodology necessary.

However, concluding from these results that the best ESD device would be a device with wide fins is not correct.

For narrow fin devices, VT is determined by the workfunction of the metal gate since the full depletion makes tuning of VT with implantation very difficult.

As a consequence, such a wide fin device can meet leakage requirements with a smaller Lg which directly results in reduced Vh and Ron.

Since these devices fail non-uniformly due to isothermal current instability at high current levels, constant current density simulations are used instead of constant power.

Obviously, forcing the same amount of current through a thinner silicon film leads to increased heating.

The thicker TBOX clearly prevents heat transfer to the substrate and hence causes increased heating of the silicon fins.

Images